AI is transforming creative industries, but few people agree on how to actually use it. Is it a shortcut? A threat? Or a tool that can extend human imagination? Two South African voices, Khumo Tsie, creative director at KN Studios, and Rachel van Rooyen, animator and lecturer at Wits University say it’s all about process. Together they offer a grounded, African perspective on how to really work with AI. For Khumo and Rachel, learning how to work with AI starts long before a prompt is typed.
Here are their practical steps for getting the most out of AI in the creative design process.
Start with the brief, not the prompt
“Everything begins with clarity,” says Khumo. Before opening a programme or typing a single prompt, he insists on a tailored client brief. Without this, AI becomes a generator of randomness rather than a collaborator. Rachel backs this up in her teaching: “You can’t outsource thinking. Students still need to learn why a composition works before they try to prompt an AI to do it.”
Draw it first
The second step may sound old-school: paper and pencil. “It’s faster, cheaper, and forces your brain to catch up with your hand,” Khumo explains. Even rough sketches narrow the scope before expensive production begins. Rachel’s ethos is similar: “Your ideas matter. AI should enhance, not replace the messy human work of figuring things out.”
Use AI for pre-visualisation
Once ideas are sketched, Khumo introduces AI to “rough out” scenes. This isn’t final art but rather considered by industry pioneers as rapid prototyping. AI outputs help spot blind spots and generate variations to test ideas before investing hours in production. Rachel adds: “Instead of accepting what the machine gives you. You’ll need to co-create with the it by learning to direct it, asking questions like: does this composition really work?”
Refine the idea with human hands
Here’s where many creatives go wrong: letting AI take over. Khumo is convinced: “AI can’t do the heavy lifting. Refinement such as nuance, cultural cues, the human feel, must come from the artist.” Rachel frames it as ethics: “Submitting fully AI-generated work to a client without disclosure undermines both artist and industry. The human touch isn’t optional, it’s what gives your work integrity.”
Let AI polish, not produce
The final stage is compositing and polish: subtle colour grading, background clean-ups, or small adjustments. “That’s where AI saves time,” Khumo says, “freeing us from repetitive tasks so we can focus on storytelling.” But he warns: “If you let AI finalise, you get mediocrity. The magic is still human.”
Rachel’s research into animation and AI highlights broader challenges: bias in training data, copyright grey areas, and the risk of eroding artistic confidence. “South African creators need to be especially alert,” she notes, “because most AI systems aren’t trained on African references or datasets. Our task is to bring African-centred storytelling into these tools.” For Khumo, the point is simple: “Change is the only constant. AI is like the camera to the portrait painter, it won’t replace the craft, but it changes the landscape. Artists who learn to co-create with it will stay ahead.” For Rachel, it’s about equipping the next generation: “As a lecturer, my students’ confidence and creativity must always come first, so we set clear boundaries on AI’s use, empowering them to enter the industry as creators, not imitators.”
Artificial Intelligence: A New Frontier for the Audiovisual Sector is supported by the French Ministry for Europe and Foreign Affairs and the Institut Français, as part of a broader strategy to support international exchange and the export of Cultural and Creative Industries.